Are You Moved by Your Social Network Application?

1. Are You Moved by Your Social Network Application? Gregory Peaker

2. Overview • This Research Paper is very well written • Good introduction in the Title • Outline: • Introduction • Reasons for work and related work • Properties of Nodes • Conclusions • Relevant Today

3. Introduction • Compare social graph containing friends vs. contact graph that is temporal network created by opportunistic contact • Prove most properties of nodes, links, and paths correlate among social and contact graphs • Describe how structure of social graph helps build forwarding paths in contact graph, allow two nodes communicate over time using opportunistic contact and intermediate nodes

4. Introduction • Study relation between social interactions and physical meetings – remains largely unexplored • 28 participants in ACM CoNEXT 2007 • Ask each participant to friend others in list of attendees • Ignore or add friend when within Bluetooth distance • Conference is reasonable group size and can be reproduced • Applied to delay tolerant network

5. Related Work • Properties of paths built in quickly varying graph is new topic • Opportunistic forwarding should be aware of social properties • Compare initial social network vs. opportunistic contacts • 9024 opportunistic contacts made

6. Results • Properties of paths built in quickly varying graph is new topic • Opportunistic forwarding should be aware of social properties • Compare initial social network vs. opportunistic contacts • 9024 opportunistic contacts made

7. Results Figure 1

8. Results Table 1

9. Results Figure 2

10. Results • The median inter-contact time grows from 6 minutes between two friends, to nearly an hour (ten time more) when nodes have distance three or four ins social graph • 75% of contacts with friends are longer than 10 minutes. Where 75% of contact with nodes distance four are shorter than 13 minutes.

11. Applications and Future Work • Tested rules for sending packets • neighbor(k): (u → v) is allowed if and only if u and v are within distance k in the social graph. • destination-neighbor(k): (u → v) is allowed if and only if v is within distance k of d. • non-decreasing-centrality: (u → v) is allowed if and only if C(u) ≤ C(v). • non-increasing-distance: (u → v) is allowed if and only if the social distance from v to d is no more than the one from u to d.

12. Applications and Future Work • Neighbor rule performs as well as most other rules. Performs significantly better than random choice. • Rule based on centrality outperforms all rules tested (reaching more than 95% of success with half the pairs) • The combination of neighbor and centrality rules naturally improves selectivity, offering more flexibility and achieve some best trade-offs

13. Applications and Future Work • Without infrastructure, must exchange information an social network allows this • Future work wish approximate centrality of a node in distributed algorithm • Centrality creates issue with targeting same set of nodes, fix this